技術標籤：python資料分析人工智慧python

KNN 演算法的原理及實現—手動寫法

import numpy as np 
import matplotlib.pyplot as plt
from math import sqrt
from collections import Counter

x = [
    [2.40,2.0],
    [4.1,2.8],
    [1.2,1.4],
    [0.6,3.7],
    [1.3,2.5],
    [6.4,4.2],
    [1.7,4.5],
    [8.2,3.5]
]

y= [0,0,0,0,0,1,1,1]
sample =
 
 [5.0,6.2]
k = 3

x_train = np.array(x)
y_train = np.array(y)

distance =[]
for i in x_train:
    d =sqrt(np.sum(i - sample)**2)
    distance.append(d)
nearst = np.argsort(distance)[:k]
# print(nearst)
k_y = [y_train[j] for j in nearst]
# print(k_y)
count = Counter(k_y)
print(count)
count.most_common
 
(1)[0][0]

plt.scatter(x_train[y_train == 0,0],x_train[y_train == 0,1],color = 'g')
plt.scatter(x_train[y_train == 1,0],x_train[y_train == 1,1],color = 'r')
plt.scatter(5.0,6.2,color = 'b')
plt.show()

用 KNN 演算法，對 scikit-learn 內建的手寫數字識別資料集進行預測

1.匯入scikit-learn 內建的手寫數字識別資料集
import numpy as np
import pandas as
 
 pd
from sklearn import datasets
iris = datasets.load_digits()

# iris.keys()
x = iris.data
y = iris.target

2.將資料分為訓練集和測試集
from sklearn.model_selection import train_test_split
x_train,x_test,y_train,y_test = train_test_split(x,y,test_size = 0.2,random_state = 666)

3.將資料進行歸一化
from sklearn.preprocessing import StandardScaler
standscaler = StandardScaler() # 建立例項
standscaler.fit(x_train) # 訓練資料，不需要特徵y   儲存計算出來的均值和方差

# 對資料及逆行轉換，用於測試資料
std_x_train = standscaler.transform(x_train)  # 用fit中儲存的均值和方差來轉換x_train，使x_train標準化
std_x_test = standscaler.transform(x_test)

4.構建KNN模型
from sklearn.neighbors import KNeighborsClassifier
clf = KNeighborsClassifier(n_neighbors = 6)
clf.fit(std_x_train,y_train)
clf.score(std_x_test,y_test)

5.對KNN模型進行網格搜尋，尋找最優的模型和引數
from sklearn.model_selection import GridSearchCV
clf = KNeighborsClassifier()
param_grid = [{
    'n_neighbors':[i for i in range(0,10)],
    'weights':['uniform','distance']
}]
gs_clf = GridSearchCV(clf,param_grid = param_grid)
gs_clf.fit(std_x_train, y_train)
gs_clf.best_estimator_
gs_clf.best_params_
# gs_clf.best_score_
6.利用最優的模型對測試集進行預測，計算預測的準確度
clf = KNeighborsClassifier(n_neighbors = 6,weights = 'distance')
clf.fit(std_x_train,y_train)
clf.score(std_x_test,y_test)